RBMX contributes to hepatocellular carcinoma progression and sorafenib resistance by specifically binding and stabilizing BLACAT1

Patient samples and tissue samples

Tumor samples from 60 tissues and adjacent normal tissues were obtained from patients who underwent liver resection at the Liver Transplantation Center in in Seventh People’s Hospital of Shanghai and simultaneously stored frozen in liquid nitrogen within 10 minutes. Patient’s character was presented in Table 1. Identification and confirmation of these tissues were performed using histology. All patients signed the informed consent and the study was approved by Institute Research Ethics Committee of the Seventh People’s Hospital of Shanghai University of TCM.

Cell line and culture

The human normal hepatic cell LO2 and hepatocellular carcinoma cell lines (HepG2, Huh-7, SMMC7721, Hep3B, HCCLM3 and MHCC97H) were purchased from the American Type Culture Collection (ATCC, Manassas, VA, USA). The cells were cultured in RPMI-1640 medium (Gibco; Thermo Fisher Scientific, Inc., Waltham, MA, USA) with 10% fetal bovine serum (FBS) (Gibco; Thermo Fisher Scientific, Inc.) at 37°C with 5% CO₂.

Establishment of sorafenib-resistant cells

To determine the half inhibitory concentration (IC50), cells were treated with sorafenib in 96 well plates and cell viability was determined for 3 days. Cells at a concentration of 1×10⁴ cells/well were cultured in 6-well plates and incubated with sorafenib at a concentration just below their respective IC50. Further, the sorafenib concentration was gradually increased by 0.25 μM every week, thus allowing selection of resistant cells. Sorafenib-resistant cell lines were obtained after 6 months, and were termed Hep3B-SR and Huh7-SR. These cells were maintained by continuously culturing in the presence of sorafenib.

Construction of RNA expression vectors

Two shRNA sequences targeting RBMX (shRBMX-1 and shRBMX-2) were synthesized by GenePharma (GenePharma Co, Ltd, Shanghai, China) to knockdown RBMX expression and the negative control shRNAs (scr) without sequence homology to human genes were provided by the same manufacturer. Cells were transfected with 100 nM shRNA duplexes using Lipofectamine™ 2000 (Invitrogen; Thermo Fisher Scientific, Inc., Waltham, MA, USA) following manufacturer’s protocols. The mock transfected cells were transfected with Lipofectamine 2000 without shRNA. PCMV6-RBMX (RBMX) and pCMV6-BLACAT1 (BLACAT1) was synthesized and cloned into the cloning site of pCMV6 plasmid (Origene) respectively. PCMV6-Entry (vector) was as negative control.

Quantitative real-time polymerase chain reaction (qRT-PCR)

RNA isolation was performed using Trizol reagent (Invitrogen) according to manufacturer’s instructions and SuperScript First-Strand Synthesis System were used for cDNA preparation (Invitrogen). PCR was performed according to the requirements. Internal control β-actin was also used. Gene relative expression was normalized to β-actin using 2^-ΔΔCt. The experiments were performed in triplicate.

Β-actin qRT-PCR forward primer: 5’-GCTACGAGCTGCCTGACG; reverse: 5’-GGCTGGAAGAGTGCCTCA.

RBMX qRT-PCR forward primer: 5’-CAGTTCGCAGTAGCAGTGGA; reverse: 5’-TCGAGGTGGACCTCCATAAC.

BLACAT1 qRT-PCR forward primer: 5’-CAAGAGGAGCCGGCTTAGCATCT; reverse: 5’-ACGGTTCCAGTCCTCAGTCAG.

Western blot analysis

Western blotting analysis was performed using 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred onto a nitrocellulose membrane. The blots were incubated with primary antibodies overnight at 4°C. Following three washes, membranes were then incubated with secondary antibody overnight at 4°C. Signals were visualized with ECL. Antibodies were purchased as follows: ABCB1 from Santa Cruz Biotechnology (Santa Cruz, CA), LC3, RBMX, caspase3 and p62 from Cell Signaling Technology, β-actin from Sigma.

Immunohistochemistry assay

Immunohistochemistry analyses were performed as per [8]. Blocking of the samples was performed in normal goat serum with 5% BSA in TBS for 1 h at room temperature. Further, the sections were incubated with primary antibody at a dilution of 1:400 over-nights at 4°C and then washed with PBS three times. After incubation with secondary antibodies (16 h), the sections were subjected to a DAB reaction. The sections were photographed using a digitalized microscope camera (Nikon, Tokyo, Japan).

In vivo tumor xenograft model

The Huh7 cells were transfected with either scr or shRBMX; Hep3B cells were transfected with empty or RBMX overexpression vector. Matrigel (50%) were used to suspend the cells and the gels were subcutaneously implanted in the dorsal flanks of 6-week-old female nu/nu mice (1×10⁶). Sorafenib was administrated when the tumor were up to 50 mm³ in volume, at this point, the mice which transfected with same vector were randomized treat with DMSO or sorafenib (50 mg/kg/each 2 days, intraperitoneally). Further, monitoring of tumor growth was carried out using digital calipers, and the volume was calculated using the formula: tumor volume (mm³) = [width (mm)] 2 × length (mm) × 0.5]. Repeats were achieved by repeating experiments 2 individual times. Initial pilot experiments were performed to project the required number of mice per group. Tumors were collected for histopathological analysis, RNA and proteins extraction. Exclusion of animals were performed when humane end points were reached or in the event of death from procedure-related causes which did not cause differences in the numbers of experimental groups.

TUNEL assay

TUNEL staining was used to stain the tumor sections (Terminal deoxynucleotidyl transferase dUTP nick end labeling) (Roche, Shanghai, China). Cell positive for TUNEL were counted 20 fields with 200 × magnification. The apoptosis index was calculated according to the following formula: the number of apoptotic cells/total number of nucleated cells × 100%.

Cells proliferation assay

Cell proliferation was calculated using CCK8 assay according to the manufacturer’s instruction. After transfection and treatment, cells were incubated in well and add 10 ul cell count kit-8 solution. Co-incubation 2 h, 37°C, measurement of absorbance at 450 nm by spectrophotometer. Cell growth curve was further drawn with normalized values which were taken at OD 490.

Colony-formation efficiency assay

After transfection with RBMX and BLACAT1 gene expression vector, cells were seeded and incubated at 37°C in humidified incubator for 10 days. Cell colonies were stained with 0.5% Crystal Violet and the colonies were counted.

In vitro apoptosis assay

Approximately 1×10⁵ cells were suspended in 100 μl buffer, 5 μl of Annexin V and 5 μl of propidium iodide were further added. The mix was incubated for 15 min at room temperature in dark, according to the manufacturer’s instruction (BD Biosciences, San Jose, CA). Flow cytometric analysis was used to quantify the rate of apoptosis (3 repeats were performed) with the Beckman Coulter Epics Altra II cytometer (Beckman Coulter, California, USA).

RNA immunoprecipitation

Immunoprecipitation was performed with A/G-plus agarose beads in the cell lysates (Santa Cruz, Biotechnology, Inc. CA) incubated with appropriate antibodies for 4 h at 4°C. Further RNA was isolated, and RT-PCR was performed.

RNA pull-down assay

RNA pull-down assays were performed using the Pierce Magnetic RNA Protein Pull-down Kit (Thermo fisher). In brief, biotin-labeled BLACAT1 or antisense RNA was co-incubated with protein extracts and magnetic beads from Huh7 and Hep3B cells. The resulting magnetic bead-RNA-protein compounds were collected by low-speed centrifugation and then washed with a Handee spin column. Magnetic beads were boiled in sodium dodecyl sulfate buffer and subjected to Western blotting using GAPDH as a control.

mRNA Stability measurement

Huh7 or Hep3B cells transfected with scr or shRBMX-1 (shRBMX) expression vector and the stability of BLACAT1 mRNA variants were examined by treating the cells with actinomycin D (Sigma). At different time points, (0, 15, 30, 60, and 120 min) RNA was isolated qRT-PCR were performed, as described above. Half-life (t1/2) of BLACAT1 mRNA variants were calculated as the time required for each mRNA variant to reduce to 50% after actinomycin D treatment from its initial abundance at time 0 min. Half-lives (t1/2) were determined by nonlinear regression analysis.

Bioinformatics analysis

In order to get a wider range of evidence, the deregulated gene expression of RBMX was further verified by The Cancer Genome Atlas (TCGA) database (n = 371); also the survival correlation (http://ualcan.path.uab.edu/analysis.html). CatRAPID program were used for interaction prediction (http://service.tartaglialab.com/page/catrapid_ group).

Statistical analysis

The data are presented as the means ± standard deviations (SD). A two-tailed Student’s t-test and One-way analysis of variance (ANOVA) was used to evaluate the differences among more than three groups or six groups, in Graph Pad Prism 5 (Graph Pad Software, San Diego, CA, USA). Differences were considered significant for values of *P < 0.05, **P < 0.01, ***P < 0.001.

RBMX is overexpressed in HCC tissues and HCC cell lines

To confirm the expression of RBMX in tumors, transcript levels were compared between the tumor and normal tissues. Based on the evidence from TCGA databases, RBMX transcript levels were observed to be highly upregulated when compared to the normal tissues (Figure 1A), as well as the association of clinicopathological characteristics of HCC patients with RBMX expression was showed in Table 1. And that, it was also observed the survivability of patients with high expression of RBMX had poor prognosis when compared to the patients with low or medium expression of RBMX (Figure 1B). In six individual patient tumor samples western blotting analysis was performed with comparisons to the normal tissue samples (Figure 1C). It was clear that there was significantly higher expression of RBMX in the patient samples, which was also observed in the mRNA level (Figure 1D). Immunohistochemistry analysis was performed in hepatocellular carcinoma (HCC) patient tissues and consistent results were evident that RBMX was highly expressed (Figure 1E). Commercially available hepatic cell lines such as LO2, HepG2, Huh-7, SMMC7721, Hep3B, HCCLM3 and MHCC97H were checked for RBMX expression (Figure 1F, 1G). Among these, Huh-7 displayed the highest expression of RBMX followed by HepG2. Further, the sensitivity of HCC cell lines to sorafenib was detected under increasing concentration (Figure 1H). The results indicated that there is some connection between expression of RBMX and HCC cells vitality in sorafenib. Then, Huh-7 or Hep3B cells were choose for further experiments.

Crucial role of RBMX in HCC cell’s proliferation and apoptosis

To understand the effect of RBMX in HCC progression, silence or overexpression experiments were performed in two different cell lines first. RBMX gene was silenced in Huh-7 cells, and silencing efficiency was confirmed using qRT-PCR analysis and western blotting analysis, and it was evident that the shRBMX-1 and shRBMX-2 had a much lower RBMX expression compared to the non-functional control shRNA (scr) (Figure 2A). RBMX overexpression assay was performed in Hep3B cells, and overexpressing efficiency was also verified in (Figure 2B). Then, proliferation assays were performed with these cells using cell counting kit. The results indicated that RBMX silenced cells have decreased proliferation (Figure 2C), on the contrary, RBMX overexpressed cells displayed a higher proliferation ability (Figure 2D). Further, colony formation capacity shows shRBMX Huh-7 cells had lesser colony forming units when compared to scr (Figure 2E), adversely, Hep3B cell’s colony units was significant increased when overexpressing RBMX (Figure 2F), this results showing a strong positive correlation between RBMX and proliferation. TUNEL assay were performed to evaluate the cells apoptosis in RBMX overexpressed and silenced cells. It was evident that the cells which RBMX were silenced had very high number of apoptotic cells when compared to the control (Figure 2G), contrarily, the dead cells are fewer in RBMX overexpressed Hep3B cells (Figure 2H). Taken together, these data demonstrated that RBMX could facilitate HCC cells proliferation and inhibit cell apoptotic.

RBMX contributes to HCC cell lines sorafenib resistance

To further understand the role of RBMX in chemoresistance, sorafenib resistance HCC lines were established as previous (Huh7-SR, Hep3B-SR), and CCK-8 assay was performed for drug resistance evaluation. As show in Figure 3A, 3B, resistant cells show great improvement of sorafenib resistance ability. Then, the expression of RBMX was measured in sorafenib resistance cell lines. We can observe that, the level of RBMX mRNA and protein significantly increased in Huh7-SR/Hep3B-SR cells compare to Huh7/Hep3B native cell (Figure 3C, 3D). To further clarify the role of the RBMX gene in cell resistance, following the processing as parental cells, Huh7-SR and Hep3B-SR cells were transfected with shRBMX or RBMX overexpression vectors, respectively. The transfection efficiency was validated by QRT-PCR and western blots (Figure 3E, 3F). Next, CCK-8 assay was performed for evaluation of RBMX function in RBMX overexpressing and silenced cells. With increasing concentrations (0-20 M) of sorafenib, it was significantly decreased of cell relative viability of Huh7/Huh7-SR cells which RBMX silenced (Figure 3G, 3H). On the contrary, RBMX overexpressed cells, Hep3B/Hep3B-SR, had higher cell viability than empty vector group (Figure 3I, 3J), thus indicating an important sorafenib resistance role of RBMX in HCC cells. Consistent with cell viability assay, cells apoptosis level, measured by flow cytometry, was negative with RBMX expression (Figure 3K-P). For further clarification, chemoresistance and apoptotic maker genes were checked next. ABCB1 (ATP Binding Cassette Subfamily B Member 1, a vital indicator of multidrug resistance), specific indicators of drug resistance was measured with western blot in parental and SR cells. Results indicated that increased expression of RBMX did increase ABCB1 expression in Hep3B/Hep3B-SR cells (Figure 3Q). And, ABCB1 expression decreased significantly in both Huh7/Huh7-SR cells when RBMX was silenced. Simultaneously, caspase-3 was selected as apoptotic maker. From the data, it was clear that caspase-3 expression was reduced while RBMX was upregulated (Figure 3Q), but when RBMX was silenced caspase-3 expression was highly upregulated. Taken together, RBMX could positive regulate parental and SR cells proliferation and drug resistance under sorafenib processing, which indicated that RBMX plays a key role in sorafenib resistance of HCC cells.

RBMX promote tumor growth and sorafenib resistance in model

Next, the role of RBMX in tumor growth and drug resistance was studied. For this, shRBMX Huh-7 cells and RBMX up-regulated Hep3B cells were injected into the dorsal flanks of nu/nu mice with or without sorafenib treatment, respectively. The tumor volumes were subsequently measured every week using digital calipers. It was evident that the tumor size increased slowly from day 7 to day 35 in the shRBMX group of Huh7 cells injection, when compared to the scr control group at the same time points (Figure 4A). On the contrary, the tumor size significantly increased in RBMX group compare to empty vector group of Hep3B cells injection (Figure 4B). There also showed a positive effect of RBMX in HCC tumor growth under sorafenib treatment. The tumor volume increased dramatically when RBMX overexpression, even that with sorafenib treatment. But, RBMX knockdown aggrandized the therapeutic efficacy of sorafenib (Figure 4A, 4B). At day 35 the mice were sacrificed and the tumors were taken out for weight measurement and pathologic analysis. As shows in Figure 4C, it was observably that no matter what kind of cells in tumorigenesis, tumor weight was heavier in the group with higher RBMX expression. More intuitively, tumors feature were showed in Figure 4D and 4E. Further, these samples were subjected to histopathological analysis (Figure 4F). Silencing and overexpression efficiency was confirmed with RBMX immunohistochemistry staining with anti-RBMX antibody, where in the shRBMX group tumor section displayed less to no positive cells, and RBMX overexpression group showed great proportion of RBMX staining. Additionally, hematoxylin and eosin staining of the tumor sections from the control group and RBMX up-regulation group showed distinct cells with clear darker nuclei typical of cancerous cells, but shRBMX group and sorafenib treatment groups showed lesser nuclei, indicating a decrease in tumorous cells. TUNEL assay of these sections displayed more apoptotic cells in the shRBMX group compared to the scr group; and the RBMX overexpression decreased cells death under sorafenib treatment. Finally, Ki67 staining was also performed which indicated more proliferative cells present by the darker nuclei coloration in higher RBMX expression group, this shows the RBMX expression rescued the HCC cells life from sorafenib toxicity. Taken together, these data indicated that RBMX promote tumor growth and sorafenib resistance by improve cells proliferation and decrease apoptotic.

RBMX mediates sorafenib resistance by induction of autophagy

Further experiments were also performed to understand the mechanism of how RBMX participate in sorafenib resistance. Interestingly, we found that autophagic flux was abundant in hepatoma carcinoma cells with sorafenib performance; but decrease rapidly when cells pretreatment with autophagy inhibitor, bafilomycin A1 (BAF) (Figure 5A, 5B). The protein detection of LC3 also indicated that autophagy was happened in hepatoma carcinoma cells responses to sorafenib stimulation. To further confirm whether RBMX regulates sorafenib-induced cellular autophagy, RBMX deletion assays were performed in HCC cells. It was observed that the cells autophagy maker LC3 significantly increased in cells which RBMX overexpressed, but reduced when RBMX deleted (Figure 5D, 5E). Western blotting analysis indicated that the LC3 II was significantly increased and p62 was downregulated in the presence of RBMX, but when RBMX was silenced then LC3 II and p62 were significantly opposite (Figure 5F, 5G), respectively. Thus, the above-mentioned experiments specifically indicated the roles of RBMX mediate sorafenib resistance of hepatocellular carcinoma cells by activating autophagy.

RNA-binding protein RBMX binds and stabilizes BLACAT1

Many studies have showed the chemotherapeutics resistant ability of BLACAT1in multi-cancers, but the specific mechanism of action is unclear. For this concern, we study the relationship between RBMX and BLACAT1 in HCCs. First of all, BLACAT1 level was measured in HCC patient tissues (Figure 6A), also the HCC TNM stage and survival rate associate with BLACAT1 level were analyzed respectively (Figure 6B, 6C). As shown, BLACAT1 is significantly expressed and has a poor prognosis in HCC patients. Besides, BLACAT1 is overexpressed in commercial HCC lines (Figure 6D). Additionally, BLACAT1 expression is positively correlated with RBMX (Figure 6E). As mentioned previously, another study had identified the hnRNP G’s RNA-recognition motif binds to a loose consensus sequence containing a CC (A/C) motif (13). In this study, we identified many similar CCAG consensus sequence by comprehensive mutational analysis of the RBMX-binding sites in the lncBLACAT1. We identified a total of 29 such sites in BLACAT1 sequence (Figure 6F). More than that, CatRAPID online algorithm also performed for RNA-protein interaction prediction (Figure 6G). Of interest, this program revealed a higher interaction DP value, reach up to 98%.On the basis of this evidence, RNA immunoprecipitation assays were go ahead for interaction validation. RNA immunoprecipitation analysis followed by qRT-PCR showed that sample precipitated with RBMX had a high fold expression of BLACAT1 (Figure 6H, 6L). Also, the RNA pull-down shows BLACAT1 bound with RBMX (Figure 6I and 6M). Further experiments were performed to investigate the stability of BLACAT1 mRNA in the absence of RBMX. Cells were either transfected with scr or shRBMX and then treated with actinomycin D to block de novo transcription. Then, the levels of BLACAT1 were quantified using qRT-PCR and normalization was performed to 18S rRNA levels. It’s obvious that, BLACAT1 mRNA had decayed to almost 60% within 8 mins in both Huh7 and Hep3B cells when RBMX was silenced (Figure 6J, 6N). This indicated that the stably expressing of BLACAT1 is potential dependent on RBMX exist. As opposed to Figure 2A, 2B, to further verify the intracellular expression relationship between RBMX and BLACAT1, Huh7 and Hep3B cells were transfected with RBMX overexpression or knockdown vectors, respectively (Supplementary Figure 1). The expression level of BLACAT1 was distinctly increased with the overexpression of RBMX in all these cells, and when RBMX was silenced, BLACAT1 was reduced (Figure 6K, 6O). Taken together, these data indicate a strong positive interaction between RBMX and BLACAT1.

RBMX and BLACAT1 interact to co-regulate HCC cells proliferation

Overexpression and knockdown studies were further used to evaluate the effect of RBMX and BLACAT1 on cell proliferation. For this purpose, Huh7 cells of RBMX knock-down and Hep3B cells of RBMX overexpressing were transfected with BLACAT1 overexpressing and shBLACAT1 vectors, respectively. The transfection efficiency, which BLACAT1overexpression in Huh cells and BLACAT1 knockdown in Hep3B cells, were proved by QRT-PCR, MTT and colony formation assay (Supplementary Figure 2). Then, RT-PCR for BLACAT1, cell viability and colony formation assay were performed. The data shows, it’s evident that BLACAT1 is decreased (Figure 7A) in Huh7 cells when RBMX was silenced even BLACAT1 was overexpressed. As BLACAT1’s expression, the proliferation and colony formation ability of Huh7 cells also declined (Figure 7B, 7C), similarly. However, in Hep3B cells which BLACAT1 was silenced while RBMX was up-regulated, the BLACAT1’s expression (Figure 7D), cell viability (Figure 7E) and colony formation density (Figure 7F) have not been rescued by RBMX overexpression compare to Hep3B cells of BLACAT1 was silenced only. These tests also performed in sorafenib resistance cells (Huh7-SR and Hep3B-SR) with same processing. Consistent with the above results, the BLACAT1 expression level (Figure 7G) and cells proliferation ability (Figure 7H, 7I) in Huh7-SR both similarly declined. Additional, Hep3B-SR cells appear identical tendency of these indicators (Figure 7J-L). In previous experiments, we had shown that RBMX increases cell viability; here we show that BLACAT1 can also increase cell viability but only in the presence of RBMX. Hence, indicates HCC cells proliferation capacity is dependent on RBMX and BLACAT1 co-regulation.

RBMX stabilizes BLACAT1 facilitate HCC cancer cells sorafenib resistance by autophagy

To further determine whether BLACAT1 is involved in sorafenib resistance in hepatocellular carcinoma cells via RBMX, toxicity experiments were performed in Huh7 and Hep3B cells with different expression of RBMX and BLACAT1. Initially, we found that the Huh7 cells which BLACAT1 overexpressed remained higher cell viability with concentration gradually increased sorafenib processing (Figure 8A), than which the empty vector group. In opposite, the Hep3B cells viability, which BLACAT1 was silenced, was decreased quickly under sorafenib challenge compare to scr group (Figure 8D). This indicating that BLACAT1 itself could contribute to sorafenib resistance of HCC cells. Then, the effect of RBMX in BLACAT1 mediated HCC cells resistance was studied next. As the results showed (Figure 8B), contrast with BLACAT1 overexpressed cells, the Huh7 cells viability decreased significantly with sorafenib treatment when RBMX was efficiently silenced, which making the cells highly susceptible to sorafenib, even if BLACAT1 was overexpressed. But, what interesting is, RBMX up regulation cannot rescue Hep3B cells vitality when BLACAT1 was silenced (Figure 8E). This explored that RBMX is indispensable in BLACAT1 mediated HCC cells resistance. Further, the multidrug resistant marker (ABCB1) and apoptotic marker (caspase-3) were checked in these two cells by western blot. The data showed that ABCB1 expression raised and caspase3 declined clearly respectively, but the silence of RBMX reversed the expression levels of maker genes (Figure 8C). As expected, ABCB1 and caspase-3 expressions just the reverse in Hep3B cells compare to Huh 7 (Figure 8F). Additionally, cells autophagy was detected for further verification. Consistent with cell viability, we observed that BLACAT1 overexpression seems to promote LC3 expression in Huh7 cells with sorafenib processing (Figure 8G), but LC3 expression seems to be declined when RBMX was silenced even in the presence of BLACAT1. Specifically, the LC3 expression deduced observably in Hep3B cells when BLACAT1 was silenced under sorafenib treatment, however, this expression level not recovered while RBMX overexpressed. And, the intensity of LC3 was quantified in Figure 8H. This indicated that the autophagy induced by BLACAT1 for sorafenib resistance regulated via RBMX. Furthermore, western blot were performed next. The proteins extract from Huh7 and Hep3B cells, which with sorafenib processing, are measured. Coincident with foregoing data, LC3 protein level also improved when BLACAT1 up regulated in Huh7 cells, but decreased when shRBMX present (Figure 8I). The LC3 of Hep3B always downregulated when BLACAT silenced, even though RBMX present (Figure 8J). Also P62, another autophagy marker, seems to be downregulated when BLACAT1 is upregulated in Huh7 cells, but when RBMX is silenced they seem to be upregulated (Figure 8I). And, this is different in Hep3B cells (Figure 8J). Hence, silencing of RBMX significantly affect the drug resistance ability by upregulating apoptosis and downregulating autophagy. Taken together, these data indicating the RBMX is essential in BLACAT1 mediated sorafenib resistance in HCC cells.